AN ULTRAHIGH VACUUM SYSTEM by HENDRIKUS WILLEM H. VANANDEL B.Sc, University of British Columbia, 196.2 A THESIS SUBMITTED .IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF MASTER OF SCIENCE - in the Department of PHYSICS We accept this thesis as conforming to the, required standard THE UNIVERSITY OF BRITISH COLUMBIA October, 1963 . I n p r e s e n t i n g t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at. the U n i v e r s i t y of B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r reference and study. I f u r t h e r agree that per-m i s s i o n f o r extensive copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s r e p r e s e n t a t i v e s . I t i s understood that copying, or p u b l i -c a t i o n of t h i s t h e s i s f o r f i n a n c i a l gain s h a l l not be allowed without my w r i t t e n p e r m i s s i o n . Department of Physics The U n i v e r s i t y of B r i t i s h Columbia,. Vancouver 8, Canada. Date October, 1963. ABSTRACT An ultrahigh vacuum system has been constructed for the purpose of f i l l i n g discharge tubes with gases without introduction of impurities. An ultimate pressure of 6 x 10-10 mm. of Hg. has been reached before f i l l i n g . Two tubes have been constructed and f i l l e d with Neon without c r i t i c a l contamination. i i ACKNOWLEDGMENT I should like to express my thanks to Dr. R. A. Nodwell whose guidance arid encouragement has been of great help in the completion of this project; to Mr. John Lees for his many hours of glassblowing and for his helpful advice on vacuum technique; to Mr. John Turner for a l l his time devoted to the design and construction of the electronic units; to Alex Fraser and other members of the technical staff for their co-operation, and to the members of the Plasma physics group for their continued interest and assistance. v TABLE OF CONTENTS Page ABSTRACT i i ACKNOWLEDGEMENT v INTRODUCTION 1 CHAPTER I - THEORY Introduction 4 1. Mechanisms causing pressure change a. Pumping with external pumps 5 b. Backstrearning 9 c. Adsorption 10 d. Desorption 13 e. Diffusion 15 f. Permeation 17 g. Ion pumping 18 2. Pressure equation 19 CHAPTER II - APPARATUS 32 1. Pumps 32 a. Backing pump 32 . b. Diffusion pump 32 2. Valves 34 3. Gauges 35 a. McLeod gauge 35 b. Ionization gauge i 36 c. Discharge gauge 43 d. Oil manometer 47 4. Gas handling system 47 5. Oven 48 6. Electronics 49 CHAPTER III - EXPERIMENTAL PROCEDURE AND RESULTS 50 BIBLIOGRAPHY 57 - i i i -Page ILLUSTRATIONS 59 1. General arrangement of the vacuum system 59 2. D i f f u s i o n pump 60 3. B a f f l e 60 4. O i l Manometer 60 5. Neon f i l l i n g system 60 6. Valve e x t e r i o r 61 7. Graph showing leakrate versus c l o s i n g torque for a l l metal valves 61 8. Discharge gauge 62 9. Graph showing s t r i k i n g and e x t i n c t i o n voltage as a function of a i r pressure for a t y p i c a l discharge tube 62 10. C a l i b r a t i o n curve for discharge gauge 63 11. C a l i b r a t i o n curve f o r discharge gauge 64 12. Ion gauge con t r o l unit 65 13. Ion current amp l i f i e r unit 66 - i v -INTRODUCTION It is well known that the presence of small amounts of impurity greatly affects the spectroscopic emission of most gases and vapors. It is gener-ally observed that the introduction of a foreign gas may drastically reduce both the intensity of resonance radiation and the coefficient of absorption for many electronic transitions (1). The extent to which the impurity affects the gas under study depends on the amount and the nature of the impurity (2). Also, a particular impurity will affect one transition more than another. In this laboratory, experiments performed on the absorption properties of Neon showed that the coefficient of absorption from excited states is crit ically dependent on the purity level of the Neon gas. With the existing facilities for the preparation of a Neon discharge tube it was not possible to prevent serious contamination of the Neon. For this reason it was decided that an ultra - high vacuum system would be built which could be used for the preparation of gas discharge tubes. This thesis deals with the apparatus and techniques involved in the construction and operation of such a system. Some aspects of ultra - high vacuum theory are also discussed. The project described served as a link in a chain of experiments performed in this laboratory in order to measure transition probabilities in excited Neon. One possible reason for the decrease of intensity of resonance lines in the presence of a foreign gas is the depopulation of excited energy states of atoms of the major constituent because of collisions with the foreign gas atoms or molecules. Instead of giving up its energy in radiation, the excited atom transfers some or a l l of its excitation energy to the foreign gas, which carries it off as kinetic energy. The probability for this to - 1 -- 2 -occur is obviously related to the probability that a c o l l i s i o n w i l l occur between the two atoms. Hence those states with long lifetimes are more easily depopulated than those with very short lifetimes, and correspondingly the effect of the foreign gas on the transitions from very shortlived states is smaller than on those from states with a relatively long lifetime, for states with the same c o l l i s i o n cross section. The absorption properties of a gas are of course similarly affected by impurities. The effect may be particularly severe in the case of absorption taking place from a metastable state to a higher state. The metastable states have a lifetime of the order of milliseconds or more, and the proba-b i l i t y for a c o l l i s i o n to occur is very high. Hence the population density of metastable states may be lowered to such an extent by the contaminant that the absorption becomes very weak indeed. Meissner (3) reports in a paper on the absorption of excited Neon that even the smallest traces of Hydrogen present in the Neon discharge decreased the absorption drastically. Ladenburg (4) in a paper on the anomolous dispersion of excited Neon confirms this and shows that only under extremely pure conditions is one able to observe effects dependent on sufficient population of metastable states, such as absorption and anomolous dispersion involving transitions from these states. The experimental arrangement formerly used in this laboratory for absorption measurements was as follows. A discharge tube with windows on the ends was permanently attached to a conventional vacuum system served by an o i l diffusion pump. The system was pumped down to a pressure of approxi--6 ^ mately 10 mm. of Hg. after'which i t was f i l l e d with Neon to a pressure of a few millimeters. Powdered Uranium was employed as a Hydrogen getter. A potential difference of about a kilovolt was applied across the electrodes of the discharge tube to give a steady orange glow of Neon. A continuum flash background source was used and the absorption of this light by the Neon dis-charge photographed on a spectograph. Since absorption lines of excited Neon failed to appear in sufficient intensity on the spectrograph under the most varied experimental conditions, i t was suggested that although the greatest care was taken to avoid contamination of the Neon, the system used was not sufficiently pure for the measurements required. This was further confirmed by the fact that upon continued use, the color of the discharge began to deteriorate. Balmer lines of Hydrogen were also clearly visible on the spectrograph whenever a picture of the Neon discharge by i t s e l f was taken. With the use of ultrahigh vacuum techniques i t was hoped that these d i f f i c u l -ties could be avoided. ' The Neon used in this project was the purest available commercially and was reported to have the following impurities present: Helium 80 parts per million Oxygen 50 " " " Nitrogen 10 " " " Hydrogen 10 " " " Water 1 " " Hence the quality of the system used had to be such that the level of impurity did not rise significantly above the quoted values due to the transfer of the Neon to the discharge tube and i t s subsequent use. As w i l l become apparent in the rest of this thesis, this requirement necessitated the use of ultra high vacuum techniques. CHAPTER I THEORY Introduction There exist many mechanisms which may be responsible for the change in pressure in a particular system making use of the removal or introduction of matter in the gas phase. The most important ones are the following, assuming leaks have been eliminated: a. Pumping with rotary or diffusion type pumps. b. Backstreaming from the pumping units. c. Adsorption of molecules to the walls of the container. d. Desorption of molecules from the walls of the container. e. Diffusion of molecules from the interior of the walls of the container. f. Permeation of molecules through the walls of the container. g. Electrostatic entrapment of ions. There are a number of ways in which these processes may be classified; external pumping, adsorption and ion entrapment are mechanisms leading to a decrease in pressure, while the others a l l tend to increase the pressure in the system; whereas one may at any time choose to turn off the pumps, the other processes go on at a l l times and can not be simply controlled. Pumping and backstreaming always occur at the same time, and adsorption and desorption likewise. Desorption, diffusion, and permeation together are usually referred to as "outgassing". In any particular system, not a l l these processes are equally important. The relative importance of the different mechanisms generally depends on many parameters of the system, such as the pumping speed of the external pumps, the temperature, the geometry of the wall material (e.g. surface to volume ratio), - 5 -and the nature and condition of the surfaces exposed to the vacuum. Because many of the processes invariably occur at the same time i t is often d i f f i c u l t to distinguish one from the other, and therefore not much quantitative infor-mation is available which would enable one to predict the behavior of a vacuum system accurately. The d i f f i c u l t y in predicting the system's behavior is increased by the fact that no one system is identical to another as far as i t s reactions to adsorption, desorption, diffusion, and permeation is concerned. These mechanisms are always a function of the previous history of the wall material of the system. However, simple calculations based on data obtained experimentally under more or less controlled conditions can give a f a i r l y good estimate of the relative order of magnitude of the different processes in any particular experimental situation. In what follows these processes w i l l be discussed in some detail in turn. For each of the processes an expression w i l l be derived which gives the rate of change of pressure as a function of time i f only that particular process were in opera-tion. After that the processes w i l l be combined in one equation giving the' pressure as a function of time, in terms of a l l the different parameters ' connected with the separate mechanisms. It must be emphasized, however, that a l l the calculations only serve to at best give a semiquantitative picture of the behavior of a vacuum system. References can be found in any standard book on vacuum technique. (See for example (5) or (6). For the adsorption and outgassing processes, which are particularly of importance for ultrahigh vacuum technique, a good reference is (7) and also (8)). The latter two also describe the general requirements for ultrahigh vacuum. 1. Mechanisms Causing Pressure Change. a. Pumping with External Pumps. The process of pumping with an external pump is nothing more than the - 6 -coming to equilibrium of a gas which has an externally maintained pressure gradient set up in i t . Consider therefore the following idealized system. T A T X nr Enclosures I and II (with volumes and V^, pressures p^ and p£, number densities n^ and n£), characterized by temperature T, are joined by an aperture of area A. The stepfunction pressure gradient of height p£ - p^ appears at A. We assume that p£ > Pp From the Kinetic Theory, for a gas of density n molecules per cc., the number of molecules striking unit area of a wall per unit time is given by where k is Boltzmann's constant, m is the mass of one molecule, and T is the absolute Temperature. Hence we can say in the above model that the net number of molecules, v> 2 ) , coming through the aperture A per unit time in the preferred direc-tion (II to I) is given by V C D 21 We define the Conductance C of the aperture as the net number of mole-cules traveling through the aperture in the preferred direction per unit time - 7 -per unit density difference. Thus r - V A > = A M The dimensions of C are easily seen to be<-il_J We now derive an expression for C in terms of p^, P2, V^, and V^. From the elementary gas laws, (3) so that, at constant Temperature, As -defined, dl to . _ _ dv->. is the number of molecules traveling from II to I per second; hence i t is equal to the rate at which molecules leave II minus the rate at which molecules enter II; that is kT - 8 -By equation (3), the difference in densities, n2 - n^, is given by Hence, We now specialize this model to the case of pumping by defining enclosure I to be an ideal pump, having the properties that Vv = O and —& = O This defines C as the pumping speed S, and we have From this we obtain the equation for the pressure rate of change due to pumping in enclosure II, dropping the subscripts, 4fc _ s f , . d t V r The idea of pumping speed can be very naturally extended to include a l l mechanisms for the removal of gas out of or influx of gas into a system; it is then defined by the equation Q (5) where Q is a measure of the flow of gas into or out of the system; the; dimensions of Q are pV per unit time. In this thesis we shall use the rather convenient pV unit of mm. of Hg. - l i t e r s , abbreviated mm. - l i t e r s . Thus an influx of ga's Q of 10 mm. - l i t e r s per second means that for a one l i t e r system the pressure w i l l rise 10 mm. in one second. This quantity Q is par-ticularly useful i f the rate of gas influx into a system is constant. We note from this derivation that the rate of evacuation for any type of pump involving the kinetic flow of gas through an ori f i c e is directly propor-tional to the quantity / • This means, for example, that Hydrogen is pumped five times as fast as Nitrogen. We also see that the minimum cross-sectional area of the system opening on the pumping side sets an upper limit to the pumping speed of the system; for Oxygen at room temperature this value is approximately 1 1 liters/sec. ; A n i r i l in cm2 Equation (4) is of course only true for an ideal pump. We have neglected the drag introduced by the walls of the tubing to the pump, and also the back-streaming which is always present to some extent in every pump. The f i r s t effect is not so important for our purposes; this merely changes the effective pumping speed at any particular point in the system. The second effect w i l l be discussed in the next section. b. Backstreaming from External Pumps. No practical pumps satisfy the defining requirement of the ideal pump, viz. \>> zsr O , and - T ^ 1 = O . In practice, a steady flow of gas is present going from the pump to the system. This flow.is generally constant in time and independent of the pressure in the system. We must consequently rewrite equation (4) as follows: ^ - b + O k (6) - 10 -Where Q. ^ is the quantity of gas flowing back in units of pV/t. An equivalent way of expressing this is by writing a -Here p„ must be defined as a n d c a n be seen to be the ultimate pressure s attainable with any particular pumping speed S and backstreaming influx Q^ . Once again we emphasize that this is only applicable to a system in which one may ignore a l l other effects (e.g. mechanisms c-f, page 1). Often this is not the case in practice, and the ultimate pressure li e s considerably higher than p u. c. Adsorption. One Of the most important phenomena occuring in ultrahigh vacuum systems is adsorption of gases to the walls of the system. Generally the distinction is made between physical adsorption, where gas molecules are held to the wall material by relatively weak Van der Waals type forces, and chemical adsorp-tion, where molecules combine chemically with the wall material. The latter process is often accompanied by dissociation of gas molecules, and the bonds between the gas and wall material are usually much stronger than those of physical adsorption. In this thesis we shall treat physical and chemical adsorption in the same general way, although s t r i c t l y speaking i t is not correct to do so, since as mentioned, chemical adsorption is often a more than one step process. Since this would involve us in too many details and special cases at the expense of cla r i t y , and since the assumption is not c r i t i c a l , we shall assume that physical and chemical adsorption occur in the same manner, the only difference between them being the energy of adsorption. The assump-tion appears justified for calculating orders of magnitude. - 11 -The pumping action due to adsorption depends on three things in general: - Pressure. (This determines the rate at which molecules strike the wall.) Surface area of the wall. Sticking probability. (The probability that on striking the wall a molecule w i l l be adsorbed.) We can thus write for the number of molecules adsorbed per second, =IN- = c A w d Where c is the sticking probability, A is the available surface area, and v ( p ) = Y\(p) J. kT (See equation (1) ) 2 i r m is the number of molecules striking a unit area per unit time. Hence, using equation ( 3 ) , we obtain, = _ k j dN = _ kT A N A d T t V dt V ^ It is convenient to express V in terms of the pressure; using the gas law we obtain and therefore v f t f IW-. - 12 -The quantity cA / is a pumping speed, (c.f. equation (4))which we c a l l S . Although i t s form is quite similar to the pumping speed derived for external pumps, i t differs from the latter in two important respects. First of a l l , the area A here refers to the effective surface area of the wall material, whereas for the external pump A stands for the smallest crossec-tional area of or i f i c e to the pump. The former is usually several orders of magnitude larger than the latter. The>other difference is the sticking probability factor c. This quantity is d i f f i c u l t to determine for a particu-lar system. It depends f i r s t of a l l on the gas-wall material combination. Furthermore, as can be expected for an adsorption process, i t depends on the surface coverage of the wall material. Measurements have been made by Foner et a l . (9), Schafer and Gerstacker, (10), and Becker (11). The latter deals specifically with chemical adsorption. The results can be generalized by saying that for a clean surface c is of the order of 1 for most combinations, varies directly as the percentage of the surface not yet covered during the formation of the f i r s t monolayer, and drops rather rapidly for second and higher order monolayers. For a clean surface i t can be seen that S a repre-sents a very formidable pumping speed. However, systems being pumped down from atmospheric pressure do not have clean surfaces in the above sense, and the pumping action is therefore of l i t t l e or no significance. In most unbaked systems, on the other hand, as we shall see in the next section, the reverse process of desorption is then more prevalent. In a baked system adsorption pumping can play a very important role. This w i l l be discussed in more detail later in connection with the other processes. To give an idea of the numbers involved in adsorption, the following table has been prepared. We assume that in an equilibrium situation at any pressure a monolayer of gas is adsorbed; this corresponds to a surface density - 13 -of molecules of the order of 5 x l O l 4 molecules per cm.^ . The last column then gives the ratio of molecules in the adsorbed phase (Na) to those in the gas phase (Ng), for a spherical container. The large ratios at low pressures should serve to convince any sceptic of the importance of the surface effects in ultrahigh vacuum technology. If one were to liberate one monolayer of gas TABLE I p mm. Hg Ng mol./cc. N mol./cm^ * a / N g 1 3 3 x 10l 6 5 x 1 0 l 4 7. 5 x 10"3 IO"6 3 3 x 10l° 5 x 10 i 4 7. 5.x 10 3 l o - n 3 3 x 10 5 5 x 10 1 4 7. 5 x 10 8 from the surface of a one l i t e r sphere into a perfect vacuum, the pressure would rise to 7.5 x 10"3 torr (1 torr = 1 mm. of Hg.). It is therefore of great importance to study the process of desorption as well as adsorption. This w i l l be done next. d. Desorption. As indicated above, the process of adsorption is always accompanied by spontaneous desorption of molecules from the surface of the wall material. The average time that an adsorbed molecule remains on a surface is given approximately by where E^ is the energy of activation, corresponding to the gas-wall material combination, T is the absolute temperature, and t Q is the period of thermal oscillation of the adsorbed molecule- normal to the surface. Normally t Q is about 10"I 3 seconds. It follows from the above equation that the rate at - 14 -which molecules leave the surface on the average is given by d _ N 0 - t NUC-t) -(-id) - e . RT (10) where N a(t) is the number of molecules per square centimeter adsorbed to the surface at any time t. In terms of changes in pressure, we obtain 4£ ^>Tf A ^ r k T ^ M A "Ir (ID d t V It should be noted at this point that the rate of rise in pressure due to desorption depends exponentially on the quantity E E are constants associated with the g a s - s o l i d combination. Todd (13) gives f o r d i f f u s i o n of water vapor out of b o r o s i l i c a t e glass the values -5 O mm.W E - 9020 c a l / m o l e /Gf*\JS*c Corresponding to the d i f f u s i o n process, the rate of r i s e i n pressure i s given by the r e l a t i o n Again an order of magnitude c a l c u l a t i o n w i l l show the importance of the tempe-rature i n the d i f f u s i o n process. Suppose an unbaked system i s pumped down ."for 4 hours and closed o f f from the pumps. The pressure r i s e at room tempe-rature during the next 4 hours due to d i f f u s i o n using the above equations would correspond to 2 x 10"^ t o r r . We assume a one l i t e r s p h e r i c a l system. Now suppose that while pumping the system down we had baked the e n t i r e system at 750° K. for 4 hours, and upon cooling had closed the system o f f from the pumps.' During the next 4 hours, the pressure would r i s e by only 2 x 10-10 t o r r due to d i f f u s i o n . The fig u r e s speak f o r themselves. Because the slope of the dP versus curve i s so much steeper at 750° K. than at 300° K., dt 4 hours of out gassing at baking temperature has the same e f f e c t as approxi-8 mately 10 hours at room temperature. This corresponds to about 1000 years! Another aspect of d i f f u s i o n worth considering i s the f a c t that i t goes on f o r -1/2 a very long time; Todd (13) reports that the diffusion follows the t ' law for the order of a year even at 800 K. f. Permeation. The permeation of gases through solids is of fundamental importance in that i t usually sets the limit oh the vacuum obtainable for any particular system. The phenomenon has been extensively studied by Norton (15). The process involves several steps and i$ a combination of the last three mecha-nisms discussed (adsorption, desorption, and diffusion). Atmospheric gases are adsorbed to the exterior surface of the walls where they dissolve into the wall material. They are then diffused through the wall material, and subsequently desorbed into the vacuum. In some cases the gas dissociates on adsorption (e.g. Hydrogen permeation through steel), and the permeation then takes place in the atomic state. In such cases the permeation rate varies as the square root of the pressure difference. In the case of permeation through glass, no dissociation generally takes place. The permeation rate is then given by the emperical relation, kp is a constant depending both on the material of the walls and the tempe-rature. It varies with temperature according to the relation k, = C e * T < 1 6 ) where Q is the Heat of permeation. Equation (15) is analogous to the well known heat transfer equation; p^ is the partial pressure outside the system (atmospheric) while P2 is the vacuum pressure, A the surface area, and d the - 18 -thickness of. the wall material. In normal situations, p£ <.<. Pp while is constant. Hence the rate of influx is constant1 to a very good approxi-mation. We have a corresponding rate of change of pressure given by 5v V 5V - K P v ^ < 1 7 ) It turns out that of the atmospheric gases, Helium permeates most rapid-ly through glass. This is not only borne out by the measurements of Norton (15) but Alpert and Buritz showed in a very interesting experiment (16) that the ultimate lower limit on the pressure in a ultrahigh vacuum system is set by the permeation rate of Helium through glass, which they measured to be approximately 5 x 10-13 mm. liters/sec. This is s t i l l a very small rate, but important for very high vacuum work. g. Ion pumping. Ion pumping in a vacuum system i s achieved in the following manner. Electrons from a source are caused to accelerates5M.n an electric f i e l d until they have enough energy to ionize atoms and, molecules. The positive ions thus formed are then collected on a negatively charged electrode. Special pumps have been designed u t i l i z i n g ion pumping, but they were not used in this project. However, even an ionization gauge acts like a pump because i t s operation depends on the collection of ions on a negatively charged electrode. The construction and operation of the ionization gauge w i l l be discussed in detail later, and therefore we shall not go into the details of the pumping mechanism now. The decrease in pressure due to the ion pumping can be described by an equation very similar to equation (4) (18) - 19 -where is the pumping speed due to ion pumping. An expression for in terms of the parameters of the gauge w i l l be derived later. It should be noted that there is l i t t l e backstreaming in an ion pumping arrangement such as the ionization gauge. Hence the gauge lowers the pressure until the out-gassing rate of the system is equal to the pumping rate of the gauge. This is of particular importance in well baked systems for which the outgassing rate is small. 2. Pressure Equation. Having briefly discussed each of the important mechanisms in the evacuation process, we shall now make a quantitative estimation of a vacuum system's behavior, by combining these mechanisms into one set of equations. We obtain the following: R T A summary of the symbols used is given below. V is the volume of the system. S is the pumping speed of the external pumps. - 20 -S a is the pumping speed due to adsorption. is the pumping speed of the ion pumps. A is the area exposed to the vacuum. tQ is the period of oscillation of adsorbed molecules in a direction normal to the surface. N a(t) is the number of adsorbed molecules per cm.2 of surface area, k is Boltzmann's constant. T is the absolute temperature, m is the mass of one molecule of gas. is the energy of adsorption, kd is the constant associated with diffusion (equation 13, 13a.) kp is the constant associated with permeation (equation 16.) c is the sticking probability (equation 7.) P a t is the partial pressure outside of the system of the permeating gas. d is the thickness of the wall material. Qb is the backstreaming rate. Equations (19) cannot be solved simply as they stand; i f Na(t) is eliminated, we obtain a second order equation in p which is nonlinear. The nonlinearity is the result of the fact that the pumping action of the walls is dependent on both the pressure and the number of atoms already adsorbed. The importance of this process varies with the condition of the vacuum. It cannot be neg-lected when in some way the wall surfaces are made free of adsorbed gas for a short period of time and then left to adsorb molecules from the gas phase of the system. This occurs when the system is heated to some temperature above room temperature and then left to cool. Later considerations will bear this out. However, when the system is pumped down from atmospheric pressure and has not been baked, the walls are not likely to do any pumping because many - 21 -monolayers of gas are already adsorbed and the sticking probability c is therefore small. Hence in such a case i t is justifiable to neglect the terms associated with adsorption pumping. This w i l l be done in our calculation, which applies to pumping on a system starting from atmospheric pressure. We shall also neglect the f % dependence in the diffusion term, and assume the diffusion to be constant in time. The reason for this is that a solution in closed form can not be obtained i f the t~h dependence is l e f t in. The impor-tance of this term is not in the time dependence but rather the temperature dependence, as is borne out by the sample calculation on page (\6). Hence we write the following equations instead of equation (19) j& _ hLW . - T & <») Solving (21) for N a(t), we obtain a) = N . ( O ) jut* (22) We define the following quantities: OC - i ( s + s t ) (3 . A k T : ^ g > 22 -- i d We then obtain, substituting (22) in (20), using the above definition, We now solve this equation with the i n i t i a l condition Mo} = |po o f t We rearrange equation (23) and multiply by *L . This gives Then So that - 23 -Hence Hence the solution is As can be seen from this equation, the time independent part , _X _ determines the ultimate pressure. ^ is a measure of the influx of gas due to diffusion and permeation, while (X is a measure of the pumping speed of the system. ^ can take on many different values depending on the material of the vacuum chamber. In the system used for this project, the wall material was mostly glass, although metal valves were used. For the purposes of calculation we shall assume that we have a one l i t e r glass system of surface area 1000 cm^ . Later we shall see how the presence of metal parts changes the situation. We shall also assume that we are pumping on this system with a pumping speed of one l i t e r per second at a temperature of 300° K (approximately room temperature). We f i r s t determine - 24 -From Todd's measurements (13) on Pyrex glass, we have i n i t i a l l y , (t = 1 sec) • = I - 4 -. x i o " U mm, l \ l e Y i / c ^ \ i t c . ( 3 0 0 ° K ) According to experiments of Alpert, Buritz, and Rogers (17), the major influx of gas due to permeation of glass is in the form of atmospheric Helium, and from measurements of Norton (15) we calculate — 9 x \o m m Ute.r3 / c m ^ c for unit pressure difference and thickness 1 mm. Again, is very much a parameter of the system used; here we shall assume that is very small compared to the diffusion and permeation influxes. In practice this is not true, certainly not at room temperature, unless very efficient traps are used. However, assuming to be small does show up other limitations of the vacuum system that are not so obvious. The partial pressure of Helium in the atmospher is 5 x 10-3 mm. so we get Y = ( l . 4 . 1 o - , l + 4 , 5 < i o ' ' 5 ) v m M ' 1 / ^ For a one l i t e r system with surface area 1000 cm2 then, we obtain The permeation rate is much smaller then the diffusion rate for an un-baked system and can be neglected at this stage. Once the system has been baked, however, the diffusion rate is much smaller (see page ), and then - 25 -the permeation rate is much more important. V Since J L is the pressure at t = cO , we shall c a l l i t p the ultimate pressure. Since 0( •» 1 for our sample system, we have p u •- 1.4 x IO"8 mm. of Hg. Several points should be noted at this stage. First of a l l , no mention was made of any nongaseous contaminants which may be on the glass, such as grease or o i l films deposited while the glass was being handled. These con-taminants may have a very high vapor pressure compared to the value of p u quoted above, and set a corresponding limit on the pressure that can be achieved. Considerable amounts of gas may also be trapped in the contaminat-ing film and the release of this gas, which is not necessarily governed by one of the described processes, can cause the pressure to stay high for con-siderable length of time. Our figures are therefore applicable only to systems which are not contaminated beyond having been exposed to the atmos-phere for some time. A second point is that while p u sets a limit on the pressure that can be achieved, we do not know how long i t takes to reach this pressure until we have evaluated the time dependent part of equation (24). This may in fact be a very long time, depending on the values of and X . In order to give an idea of the time dependence of the pressure in the system, a table has been prepared giving the value of the pressure as predicted by equation (24) at times t = 15 minutes, t = 24 hours and t = 10 days for the sample system used above. The parameter which is varied is E^, the energy of desorption. The reason for doing this is that E d varies for different gas-solid combinations. Measured values range from 20 cal. per mole (the heat of vaporization of liquid Helium) to several hundred thousand cal. per mole (e.g. - 26 -the activation energy of oxygen on Titanium is 236 kcal. per mole.) Not much quantitative data is available for glass-gas systems. Tuzi and Gkamoto (18) report E d for water on glass in high vacuum apparatus to be between 13 and 40 kcal. per mole. Because the values of E d for the various gases and vapors in the system are not well known, we have calculated the pressure time dependence for various values of E d, in order to see for which values of E d the desorp-tion process is important from the point of view of reaching high vacuum quickly. One d i f f i c u l t y in computing the values of the table is the choice of a suitable i n i t i a l value of Na, the number of molecules adsorbed. Briggs (19) gives a figure for water vapor on glass of the order of 5 x 10^ molecules/cm2 while the number of Nitrogen molecules per cm2 is given as 5 x IO* 4 mol./cm2-This is in-agreement with observations by Todd (13), who reports that 997. of the gas desorbed from glass on heating is water. Hence we have taken Na(0) to be 5 x 10*6. The table is given on page (27). The results of the calculations show that i f our assumptions are correct, there exists a definite range of desorption energies for which the outgassing process impedes the speedy production of high vacuum. This range li e s between 20 and 30 kcal. per mole for the p u calculated above. If the p u happens to be lower, this range is correspondingly extended. Qualitatively, these results mean that for low energies of desorption, the molecules are pumped off the walls in a very short time because they are bound by very weak forces. For high energies, on the other hand, the molecules are so tightly bound that no' appreciable desorption takes place. It is for the middle range of energies that desorption takes place at a rate which keeps the vacuum of low quality for long times. It is li k e l y that water vapor and other active atmospheric Cont'd on page 28. - 27 -TABLE (III) P (t -(mm. - 15 min.) of Hg.) P (t (mm. = 24 hrs.) of Hg.) P (t )mm. = 10 days) of Hg.) 1 kcal./mole Pu Pu p u 10 kcal,. /mole Pu Pu Pu 20 kcal./mole Pu Pu Pu 21 kcal./mole 1.8 x 10-4 Pu Pu 22 kcal./mole 4.0 x lO-^ Pu Pu 23 kcal./mole 2.0 x IO" 4 Pu Pu 24 kcal./mole 6.0 x IO"5 1.1 x IO"6 Pu 25 kcal./mole 1.5 x 10"5 6.0 x 10"6 Pu 26 kcal./mole 2.4 x 10"6 2.0 x IO"6 4. 1 x 10"7 27 kcal./mole 4.5 x 10-7 4.5 x 10"7 3. 3 x 10"7 28 kcal./mole 8.9 x IO" 8 8.9 x IO' 8 8. 9 x.,10-8 29 kcal./mole 2.5 x IO"8 2.5 x IO"8 2. 5 x IO"8 30 kcal./mole 1.7 x 10-8 1.7 x 10-8 1. 7 x IO' 8 31 kcal./mole Pu Pu Pu Table showing the time dependence of the pressure as a function of E - 53 -of Hg. Due to possible errors in calibration this value could be out by a factor of two either way, but certainly no more. It is possible to spoil the vacuum temporarily or permanently in many ways. Outgassing of the grid for too long a period or at too high a tempera-ture w i l l liberate material from the grid i t s e l f which w i l l raise the pres-sure. The best way to outgas the grid is to do i t for short periods of time (say five minutes) with the valve to the pumps open and to close this valve as soon as the grid has cooled down. In this way the grid surface has no time to adsorb any backstreaming o i l vapor from the pumps. Leaving the system exposed to the pumps for long periods of time after i t has cooled also deteriorates the vacuum. For this reason the system was never l e f t to cool off overnight, in spite of the fact that the cooling process is quite lengthy. After the lowest possible pressures had been reached, Neon was intro-duced into the system with the valve to the pumps closed. The Neon could be let into the system by breaking the breakseal on the bottle (see figure ( 5 ) ). Usually valve 4 was closed before the seal was broken; in this way the Neon would only expand into the small volume IV. By opening valve 4 very care-fu l l y , the Neon could be leaked into the other parts of the system very slowly. After enough Neon had been leaked in to maintain a discharge on the large discharge tube, the electrodes of the tube were outgassed by passing a larger than usual discharge current through them. This undoubtedly contami-nated the Neon, but since valve 4 was closed at this stage, this dirty Neon could be pumped off, and fresh pure Neon leaked in as before via valve 4. A striking characteristic of the Neon gas was that i t did not stick to the walls of the system like air does; even though the Neon pressures were as high as 10 mm. Hg., i t vzas always possible to reach ultrahigh vacuum again - 54 -without baking, by pumping the Neon off and outgassing the grid. The gauge which was intended for the measurement of Neon pressure after the tube had been f i l l e d had to be calibrated in Neon before i t could be used. Because the calibration had tio take place under the same purity conditions as the f i l l i n g of the tube (see apparatus description page (16)), this posed a problem, since any other gauge used for.calibration would introduce impurities. It was fi n a l l y suggested that the minimum amount of impurity would be introduced i f an o i l manometer were used, f i l l e d with very low vapor pressure o i l , and connected to the system via a tube f i l l e d with Zi o l i t e pellets. After the o i l manometer was put on the performance of the vacuum system changed slightly. The lowest pressure ever attained with the o i l manometer in the system was of the order of 1 x 10**^ mm. Hg. In general the system performed quite well in spite of the presence of the o i l . The manometer was placed outside the oven area, which meant that a part of the system was not baked. The calibration curves for the discharge gauge are given in figures (11,10). The calibration was found to change slightly i f the Neon were l e f t in the oil-containing part of the system for any length of time (say over-night). Therefore i t was deemed best to measure the pressure of the Neon as soon as possible after the leaking in, and to seal the tube immediately afterwards. In this way the contamination of the Neon was kept to a minimum. The second tube of Neon was prepared after calibration of the discharge gauge with the o i l manometer s t i l l in the system, to provide a check on the pres-sure. The pressure at which the tube was sealed off was 1.3 mm. Hg. This was the optimum pressure for the absorption experiment (see Ladenburg (4) ). For both tubes prepared, the pressure before f i l l i n g was in the 10-9 range. The tubes prepared were both used in the absorption experiment and showed much improvement over the system used before. The absorption lines, which were nearly nonexistent with the old system, showed up stronger and also in greater number. The H ^ line was barely visible when emission spectra of the Neon were taken, indicating that traces of Hydrogen were s t i l l present. However, i t was very much weaker for the prepared tubes than i t had ever been under similar conditions with the old system, indicating that the Hydrogen content of the discharge tube was much less than before. This Hydrogen is probably that which was in the Neon i n i t i a l l y (see page ( 3 ) ). The experiments with the absorption tubes are s t i l l in progress, and there-fore i t cannot be said with certainty at this stage whether the Neon was pure enough for our purposes. Indications are, however, that this is the case. i There are, of course, ways in which the system can be improved. It seems likely that with careful experimentation the ultimate pressure can be lowered somewhat, particularly when the o i l manometer is removed again. This mano-meter was le f t on for the time being to check the calibration of the dis-charge gauge at some later time. If in the preparation of more discharge tubes i t appears that the limitation on the purity is not in the ultimate pressure attainable, but rather in the manufactured Neon bottle i t s e l f , various methods of further purification could be built into the system. One such method which seems particularly promising is purification by cataphoresis (28) . The removal of Hydrogen u t i l i z i n g Uranium powder as described by Dieke (29) could also be tried. In the series of experiments carried out in this laboratory for the measurement of transition probabilities in excited gases, i t is lik e l y that many tubes of the kind described above w i l l have to be prepared, not only f i l l e d with Neon, but also with other gases. It is hoped that the system built for this project, together with the experience gained, w i l l be of some use for this purpose and in the general interest of scie n t i f i c endeavor in the future. - 57 -BIBLIOGRAPHY 1. Mitchell, A.C.G. , and Zemansky, M.W., "Resonance Radiation and Excited Atoms", Cambridge University Press, 2nd. ed. , 1961. 2. Stuart, H., Zeits. fur Physik 32, 262 (1925) 3. Meissner, K.W., Ann. d. Phys. _76, 124 (1925) 4. Ladenburg, R., Zeits. fur Physik 48, 32 (1928) 5. Dushman, S., "Scientific Foundations of Vacuum Technique", John Wiley & Sons, 1949. 6. Yarwood, J., "High Vacuum Technique", John Wiley 6c Sons, 1961. 7. Redhead, P.A., Hobson, J.P., and Kornelson, E.V., in "Advances in Electron Physics" Vol. 17, p 323, Academic Press, London and New York, 1962. 8. Alpert, D., in "Handbuch der Physik", (Flugge, ed.) Vol. 12, p 609, Springer, Berlin, 1958. 9. Foner, S.N. et a l . , J- Chem. Phys. 31_» 5 4 6 (1959) 10. Schafer, K. , and Gerstacker, H., Z. Elektrochem. 60, 874 (1956) 11. Becker, J.A. in "Structure and Properties of Solid Surfaces", (Gomer & Smith, eds.) p 459, Univ. of Chicago Press., 1952. 12. Sherwood, R.G. Phys. Rev. 12, 448 (1918) 13. Todd, B.J., J. of Appl. Phys. 26, 1238 (1955) 14. Barrer, R.M. , "Diffusion in and through Solids", Cambridge Univ. Press, London & New York, 1951. 15. Norton, F.J., J. of Appl. Phys. 28, 34 (1957) 16. Alpert, D. , andBuritz, R.S., J. of Appl. Phys. 25, 202 (1954) 17. Alpert, D. and Buritz, R.S., and Rogers, W.A., J. of Appl. Phys. 25, 868 (1954) . - 58 -18. Tuzi, Y. and Okamoto, H. , J. Phys., Soc. Japan 13, 960 (1958) 19. Briggs, L , J. of P. Chem. 9, 617 (1905) 20. Alpert, D., J. of Appl. Phys. 24, 860 (1953) 21. Biondi, M.A., Rev. of Sc. Instr. 30, 831 (1959) 22. Nottingham, W.B., J. of Appl. Phys. 8, 762 (1937) 23. Nottingham, W.B., MIT Conference on Physical Electronics (1947) 24. Alpert, D., and Bayard, R.T,, Rev. of Sc. Instr. 21, 571 (1950) 25. Lander, J.J., Rev. of Sc. Instr. 21, 672 (1950) 26. Metson, G.H., Brit. J. of Appl. Phys., 2, 46 (1951) 27. Hirsch, E.H., Rev. of Sc. Instr. 32, 1373 (1961) 28. Riesz, R., and Dieke, G.H., J. of Appl. Phys. 25, 196 (1954) 29. Dieke, G.H. and Cunningham, S.P., J. of Opt. Soc. Am. 42, 187 (1952) I baked area I l_ # 1 valve trap ion gauge # 3 valve discharge tube -Vol. I. 4J -Vol. I l l discharge gauge Vol. II 2 valve l # 4jvalye|. , i o i l manometer \ " J T Vol. IV Neon bottle FIGURE 1 General arrangement of the system (schematic) - 60 -cooling water to system pump ' FIGURE 2 Diffusion pump (schematic diagram) K '/. 77777// FIGURE # U tube o i l manometer I. I FIGURE 3 Baffle . cooling water Neon FIGURE 5 Schematic- of Neon f i l l ing system - 61 -rn~T 11 TT FIGURE 6 Valve with driver •I 16—»-| Valve with bakeout clamp Torque in f t . lbs. FIGURE 7 Graph of closing torque versus leak rate for the a l l metal valves -62.-250 M -AM/VW-FIGURE 9 Discharge gauge circuit GAUGE i '4" 3« " 4 " igauge geometry 300--Voits 1 Pressure in mm. Hg. FIGURE 9 Graph of striking and extinction voltage versus pressure - 63 -- 64 -2.ff \ z 3 A- 5 6 7" 8 9 Pressure in mm. of Hg. FIGURE 11 Calibration curve for the discharge gauge; V Q 1 3 600 V. (Neoi^ 1 " A V * 1— 1_ 470-TL j -\ooyf / J_ —-WY ' — i — \ A A A 47Q-K- I I N 1 6 2 7 ^ - B A Y A R D - A L P E R T T U B E I - T O l c M O N I T O R -V I S O V. K—"L—lc 269 ex ; 4-TL ( lOO w) • 2IO H 4 0 0 o H T FIGURE 12 Ion gauge control unit 2.00 M ION COULECTOR C K 3 I 2 A X + I S O V. REG. 3.3 K 0-2 M A METER £ 3.9 K - A / V W * — _ ISO V . 262 I I O O A -vvw*-820X1-I 4 0 l i t FIGURE 18 Ion Current amplifier